Maouene, M. 1 , Maouene, J. 2 & Canada, K. 2, 2
1 Ecole Nationale des Sciences Appliquees, Tanger
2 Grand Valley State University
The shared features that characterize the noun categories that young children first learn are a formative basis of the human category system. Recently, Hills and colleagues (2009 a,b) described the potential categorical information contained in the features of early-learned nouns by examining the binary graph-theoretic properties of developing noun-feature networks with a deterministic method: the clique percolation. The networks were built from the overlap of perceptual and functional features for words normatively acquired by children at three different ages: 20 months --21 nouns--, 25 months --56 nouns-- and 30 months -- 130 nouns--. The resulting networks had small-world structures, indicating a high degree of feature overlap in local clusters. Results also suggested that overlapping features among these nouns created higher-order groupings common to adult taxonomic designations and ad hoc categories. However, these methods are limited as they are only descriptive and yield minimal semantic information such as the degree of connectivity of local structures, whether an unspecified link of similarity exists, or identifying cliques of connectivity. To account for these limitations, we present a different type of representation, the hypernetwork (Berge, 1956), which includes semantics, and a different formalism, the Formal concept analysis (FCA), a non-deterministic method that builds relationships of containment (Wille, 1984). Further, machine-learning algorithims automatically cluster and build inclusions for the features and their nouns at the three ages mentioned above. We compare our results to the results obtained by Hills and colleagues. The power of the system lies in its automaticity and ability to form many intermediate clusters at all stages of growth of the network in addition to showing the structure's emerging paths. The results offer new and testable hypotheses on the role of shared features in the emergence of meaningful pathways within local structures, fundamental in categorizing systems.